In the 1960s Barlow developed the efficient coding hypothesis, arguing that the function of sensory encoding is to maximize the efficiency with which neural signals are transmitted. The principle of recoding [visual-sensory events into neural signals] is to find what images are expected on the basis of past experience and then to ... reserve the outputs with many impulses for the unusual or unexpected inputs. To guide visual behavior, movements of the eyes, efficiently encoded sensory signals in the striate and extra-striate cortex must be combined with stored estimates of the value, to the organism, of the actions associated with those signals. Perhaps surprisingly however, studies of the visual-saccadic system have lagged behind studies of the striate and extrastriate cortices in this regard. We know almost nothing about how reward-related saccadic signals are encoded, and this means that we have almost no understanding of how injuries and diseases that devastate brain areas ranging from the basal ganglia to the parietal cortex can be addressed by behavioral treatments. Here we propose to begin the formal study of neural coding in the saccadic system by programmatically assessing the efficiency with which the nervous system encodes information about the rewards that guide saccadic behavior in a way that will allow us to test several competing contemporary hypotheses about the function of the basal ganglia and the parietal cortex. We propose to test hypotheses about the roles of these brain areas in learning, storing, and representing the values of actions. Finally, we propose a set of causal interventions in brain activity to test the hypotheses developed by these behavioral and physiological experiments. We therefore propose to study 1) the midbrain dopamine neurons, 2) the visual-saccadic striatum (both the caudate and the ventral striatum) and 3) the lateral intra-parietal area.